Patent Application
20170126196
The present invention relates generally to circuits for providing an analog audio signal from a digital audio signal, and more particularly to audio circuits for providing such an audio signal that can be adjusted at run time.
It is now common for people to listen to music or other audio or video content using mobile devices such as smartphones or other mobile media players (collectively “smartphones”). Many users, if not most, typically listen to music through over-the-ear headphones or in-ear ear buds, although it is also possible to play music from a smartphone through small loudspeakers in the smartphone, or to connect to an external loudspeaker.
Headphones, ear buds and speakers are all analog output devices that contain one or more one or more transducers that require an analog signal from the smartphone to be activated and produce audible sound to the user. An output audio signal is sent from the smartphone by an audio subsystem in the smartphone to the transducer(s) in the chosen device to produce sound so that a user may listen to the audio output.
Smartphones, like most electronic devices in use today, store data in digital form, rather than in analog form. Thus, music, or any other audio signal is stored as a digital audio signal, and must be converted to an analog audio signal before it can be converted into sound by a transducer. (Audio data may also be streamed to a smartphone from another source over a cellular or wifi connection; this will also typically be in digital form and must similarly be converted.) For this reason, most or all smartphones contain an audio rendering system for converting digital audio data to an analog audio signal.
An audio rendering system will generally contain a digital-to-analog converter (“DAC”), and one or more amplifiers (often operational-amplifiers or “op-amps”). There may also be line driving components for driving the now-analog and amplified signal to the transducer(s). The audio rendering system takes the digital data representing an audio signal and renders it into the analog domain.
Users typically want to be able to listen to music having high sound quality. However, a transducer will reproduce any noise that is present either in the original digital audio signal or added during processing by the audio rendering system, and is thus also present in the analog audio signal. This may limit the user's perception of the quality of the sound. Thus, while unwanted noise is inevitably created in the analog domain by the process of converting digital audio data to an analog audio signal, it is desirable for audio rendering systems to have low noise.
Audio rendering systems typically have a bandwidth that extends across the entire range of frequencies considered audible to people, i.e., from 20 hertz (Hz) to 20,000 hertz, i.e., 20 kilohertz (kHz). Unwanted noise is defined as the total noise voltage in that audio band. For example, a designer might specify that the total noise voltage in the band from 20 Hz to 20 KHz should be no more than 1 microvolt (μV). Designers of audio rendering systems will specify post-conversion elements (again, after the DAC) so as to try to optimize the amount of total noise voltage present.
Since noise energy is distributed equally across the spectrum of frequencies in the band, the noise at the output depends on total bandwidth as well as the voltage. Thus, another way to describe noise is by considering it as energy that appears as a certain power per unit hertz in the band.
A state of the art low noise audio rendering system at present may, for example, have noise of 5 nanovolts (nV) per root hertz (square root of the bandwidth), i.e.:
To calculate total noise (root mean square value, or RMS), this is simply multiplied by the square root of the bandwidth. Since audio commonly has a 20 kHz bandwidth as above, the total noise of an audio rendering system with 5 nV per root hertz is:
5 nV*√{square root over (20,000)}=707 nV RMS
To improve noise in an audio rendering system, the noise in the electrical elements must be reduced, and then the above calculation performed using the total bandwidth.
One attempt at reducing noise in the prior art is known as auto mute. In auto mute, the entire channel is shut off when silence is detected. Of course, this doe not accurately represent the system in use; to test such systems when no audio signal is present, a 1 kHz tone at −60 decibels (dB) relative to full scale (i.e., full volume) is used, to prevent the auto mute from masking the real noise performance of the system. In any event, auto mute does not reduce noise when any audio signal is present.
It would be desirable to find a way to reduce noise in an audio rendering system.
A circuit and method for lowering noise in an audio rendering system is described. The described circuit allows adjustment of the bandwidth of the system during run time, so that the bandwidth is what is sufficient for the analog audio signal present at any given moment, rather than requiring the entire 20 kHz audio spectrum to be constantly present.
One embodiment discloses audio rendering device for providing an output analog audio signal from a digital audio signal, comprising: a digital to analog converter configured to convert the digital audio signal to an analog audio signal; an adjustable high pass filter configured to pass a portion of the analog audio signal above a first cutoff frequency, and to adjust the first cutoff frequency in response to a first control signal; an adjustable low pass filter configured to pass a portion of the analog audio signal below a second cutoff frequency, and to adjust the second cutoff frequency in response to a second control signal; and a controller configured to: determine any frequencies contained in the analog audio signal during a selected time interval while the analog audio signal is produced by the digital to analog converter; determine an analog signal bandwidth that is less than a full 20 kHz audio range bandwidth and that will allow the determined frequencies contained in the analog audio signal to pass through the high pass and low pass filters; determine the first and second cutoff frequencies for the high pass and low pass filters such that the cutoff frequencies define the determined bandwidth; and generate the first and second control signals to adjust the first and second cutoff frequencies in the high pass and low pass filters, so as to limit the analog audio signal to the determined bandwidth.
Another embodiment discloses a method of converting a digital audio signal to an analog audio signal, comprising the steps of: analyzing, by a controller, any audio frequencies contained in the digital audio signal during a specified time interval; converting, by a digital-to-analog converter, the digital audio signal to an analog audio signal; determining, by the controller, a bandwidth of the analog audio signal and cutoff frequencies of high pass and low pass filters sufficient to allow all of the analyzed audio frequencies to pass to an output; determining, by the controller, values for adjustable impedance components in the high pass and low pass filters that will result in the determined cutoff frequencies; causing, by the controller, switches in the adjustable impedance components to open or close to cause the adjustable impedance components to have the determined values; and passing the analog audio signal through the high pass and low pass filters to the output.
Still another embodiment discloses an audio rendering device for providing an output analog audio signal from a digital audio signal, comprising: a digital to analog converter configured to convert the digital audio signal to an analog audio signal; a first adjustable high pass filter configured to pass a portion of the analog audio signal above a first cutoff frequency, and to adjust the first cutoff frequency in response to a first control signal; a first adjustable low pass filter configured to pass a portion of the analog audio signal below a second cutoff frequency, and to adjust the second cutoff frequency in response to a second control signal; a second adjustable high pass filter configured to pass a portion of the analog audio signal above a third cutoff frequency, and to adjust the third cutoff frequency in response to a third control signal; a second adjustable low pass filter configured to pass a portion of the analog audio signal below a fourth cutoff frequency, and to adjust the second cutoff frequency in response to a fourth control signal; and a controller configured to: determine any frequencies contained in the analog audio signal in two non-overlapping portions of a 20 kHz audio range bandwidth during a selected time interval while the analog audio signal is produced by the digital to analog converter; determine a first analog signal bandwidth that is less than the full 20 kHz audio range bandwidth and that will allow the determined frequencies contained in a first portion of the analog audio signal to pass through the first high pass and first low pass filters; determine a second analog signal bandwidth that is less than the full 20 kHz audio range bandwidth and does not overlap with the first determined bandwidth, and that will allow the determined frequencies contained in the second portion of the analog audio signal to pass through the second high pass and second low pass filters; determine the first and second cutoff frequencies for the first high pass and first low pass filters such that the first and second cutoff frequencies define the first determined bandwidth; determine the third and fourth cutoff frequencies for the second high pass and second low pass filters such that the third and fourth cutoff frequencies define the second determined bandwidth; generate the first and second control signals to adjust the first and second cutoff frequencies in the first high pass and first low pass filters, so as to limit a portion of the analog audio signal to the first determined bandwidth; and generate the third and fourth control signals to adjust the third and fourth cutoff frequencies in the second high pass and second low pass filters, so as to limit another portion of the analog audio signal to the second determined bandwidth.
Described herein is a circuit and method for lowering noise in an audio rendering system. The described circuit allows adjustment of the bandwidth of the system during run time, so that the bandwidth is what is sufficient for the analog audio present signal at any given moment, rather than requiring the entire 20 kHz audio spectrum to be constantly present.
Lower noise in the audio DAC is achieved without improving the noise of each electrical element. This is based on the fact that audio is a sequence of various tones. Music and speech are constructed from tones (or phonemes in the case of speech) that do not fill the entire audio band, i.e., each element of music or speech does not fill the entire 20 kHz audio band. This observation is also the basis of the MP3 compression algorithm; in MP3 compression, it is not necessary to compress the entire data flow representing audio, but only those portions where actual audio frequencies are present.
In one embodiment, an audio rendering system, specifically the post processing elements of an audio DAC, and possibly even the audio DAC itself, can be adjusted at run time, i.e., while the audio signal is progressing or “playing,” to significantly reduce the noise.
The audio rendering system first determines the extent in frequency of the audio content; within the digital domain, as data flows to the audio DAC, an analysis is made of the spectral content of the signal, i.e., the frequencies it contains, over a certain time interval, for example, over 20 millisecond (mS) intervals. The audio rendering system may contain a controller, in addition to the DAC and amplifier(s), that may perform such an analysis.
The electronics that render the audio signal from the digital domain into the analog domain are adjusted to provide just the bandwidth required to pass the frequencies that are contained in the time interval. For example, if the signal is a tone at a frequency of 250 Hz, then the rendering electronics might be adapted to pass frequencies from say 200 Hz to 300 Hz. In the next time interval, the audio content may instead have a frequency of 3 kHz in it. The electronics are then adjusted to handle the frequencies around the more recent signal as required.
In the prior art, the audio rendering system 100 of
As above, it is believed that the current state of the art still produces noise of 5 nV per root hertz, which will result in total noise of 707 nV RMS across the 20 kHz audio bandwidth.
The second op-amp U3, resistors R2 and R3, and capacitors C1 and C4 make a low pass filter 204 which passes all frequencies below a second cutoff frequency; one of skill in the art will again appreciate how to select values of R2, R3, C1 and C4 that will result in a desired second cutoff frequency. Unlike the prior art, however, in this embodiment some or all of resistors R1 to R4 and capacitors C1 to C4 are made in such a way that they are adjustable during run time.
One of skill in the art will appreciate that audio rendering system 200 as illustrated uses what is known as a Sallen-Key architecture, and that an alternative architecture for either high pass filter 202 or low pass filter 204 may be used. Further, any of the resistors or capacitors may be adjusted to modify the RC constant for each filter, and thus adjust the cutoff frequency of that frequency; however, in practice it may be preferable to keep the resistors fixed and adjust only the capacitors, since the resistors are sources of noise while the capacitors are not, and adjusting the resistors may thus require recalculations of the noise present in the circuit.
As above, during run time, i.e., the playing of an audio signal, a controller 206 examines the digital audio signal as indicated by the arrow from the input to controller 206 and determines, for each successive time interval of a specified length (for example, 20 mS), which frequencies are presently found in that time interval. The controller 206 may then generate appropriate first and second control signals to cause adjustment of the two filters by causing adjustment of the resistors and/or capacitors for each time interval so as to set the first cutoff frequency of the high pass filter 202 and the second cutoff frequency of the low pass filter 204 to form a bandwidth adequate to pass the determined frequencies for that time interval while excluding the rest of the normal audio spectrum. Again, any of the resistors or capacitor may be adjustable, although it may be preferable to adjust only the capacitors; for simplicity, in
In the above description there is a single high pass filter 202 and a single low pass filter 204. In some situations, it will be beneficial to have more than one high pass filter and one low pass filter. For example, consider a signal with a pair of frequencies, one at 250 Hz and another at 2 kHz. In the circuit 200 above, a bandwidth from, for example, 200 Hz to 2.2 kHz might be set, which would allow both frequencies to pass while being significantly less than the full audio spectrum.
However, if there are two high pass filters and two low pass filters, then two separate bandwidths may be defined, for example, a first from 200 Hz to 300 Hz to pass the 250 Hz frequency, and a second from 1.8 kHz to 2.2 kHz to pass the 2 kHz signal. This will result in a reduction in noise that is significantly better than the already significant reduction achieved by having a single bandwidth that passes both signals.
Such a solution uses two instances of high pass filter 202 and low pass filter 204 to form two parallel channels. Controller 206 must of course be programmed to be able to determine those situations in which two channels will provide a noise advantage over a single channel, and to adjust each high pass and low pass filter appropriately. One of skill in the art will further appreciate that any number of channels may be defined where appropriate, each channel having a high pass filter and a low pass filter, and the controller programmed accordingly.
In some embodiments, such parallel channels may be used to provide a stereo signal of reduced noise, or alternatively may be completely independent. Further, in some embodiments, the process of determining which audio frequencies are “present” in an interval may deliberately and selectively ignore some frequencies. For example, suppose an audio signal contains both speech and a constant recurring sound such as that from a fan or propeller; one might regard the recurring sound as an “extraneous” and undesired input to the audio signal that generates a broadband noise. A controller might analyze the audio signal, determine that only the speech frequencies should be passed, and cause the filters to define a bandwidth (which may change for each interval as above) that passes the speech frequencies while rejecting the broadband noise due to the constant fan or propeller noise.
Any method of adjusting the values of the resistors and capacitors in circuit 200 may be used to adjust the cutoff frequencies of the high pass filter 202 and low pass filter 204. A number of such methods are described in co-pending patent application Ser. No. 15/339,617 (“the '617 application”) filed Oct. 31, 2016, and entitled Programmable Circuit Components With Recursive Architecture, which is commonly owned by the applicant of the present application and is incorporated herein in its entirety.
One prior art method described in the '617 application involves replacing a single resistor or capacitor with a plurality of resistors or capacitors, respectively, configured to be connected in series or parallel by switches which may be independently opened or closed at run time such that any or all of the plurality of resistors or capacitors may be included in the circuit, thus allowing for a plurality of possible effective values for the element replacement. A combination of multiple resistors or capacitors made in this way that has only two ends and thus appears to other circuit elements to be a single resistance or capacitance, respectively, may be thought of as a “compound element.”
For example,
It is also known to select weighted values for the capacitors to be placed in parallel as in
In other embodiments, a network of resistors or capacitors that combines both series and parallel configurations may be used. The '617 application illustrates a resistor network that includes both a series and parallel combination in a particular use case. More generally,
As is also known in the art, series or parallel combinations of devices having impedance, such as resistors, capacitors, inductors and FETs, such as those shown in
One way to obtain a uniform maximum error between a desired value and a possible value across the range of an adjustable circuit is to construct the adjustable circuit so that the possible values are logarithmically spaced equally in value, rather than linearly. As also explained in more detain in the '617 application, one way to accomplish this is to construct the adjustable compound elements in a recursive fashion.
As detailed in the '617 application, such a recursive construction allows for an adjustable element to be easily made and results in possible values distributed equally on a logarithmic scale. Further, as explained therein, in some applications such compound elements do not suffer from certain other limitations of prior art solutions, such as waster power and excessive noise.
As detailed in the '617 application, by opening or closing the switches, it is possible to select from five different values of impedance for compound element 600. However, in a simpler embodiment, only two of the possible five states of compound element 700 are used, placing elements U1 and U2 in series and U1 and U2 in parallel. Thus, a single bit control signal, called C, will suffice to select between these two states. One value of the control signal C, for example low or 0, will close switches S1 and S2 and open switch S3, resulting in U1 and U2 being in parallel. The opposite value of the control signal C, in this case high or 1, will open switches S1 and S2 and close switch S3, resulting in U1 and U2 being in series.
The connections of
Now the control signal uses 3 bits. The switches in each of the two compound elements U3 and U4 are controlled by a bit which places the simple elements U1 and U2 within U3 and U4 in either series or parallel as above. The third control signal bit controls the switches S4, S5 and S6 in the same way that the switches in each compound element U3 and U4 are controlled as shown in
This process can be repeated, so that compound element 800 may be shown as a higher level compound element U5 in circuit 900 in
Any of these types of adjustable compound elements may be used to replace the resistors R1 to R4 or capacitors C1 to C4 in the embodiment of an audio rendering system shown in
In light of the teachings herein, one of skill in the art will be able to decide how great a bandwidth should be allowed for any given frequencies detected in an interval in a particular application, and thus what the appropriate cutoff frequencies should be. Given this, it is well known in the art how to calculate the effective values of the resistors R1 to R4 and capacitors C1 to C4 that will result in any given cutoff frequencies.
The mapping of switch positions to possible effective values of the compound elements may be also be determined by the processor during run time, or alternatively the mapping of switch positions to effective values may be determined in advance and stored within the controller's programming or in a separate memory accessible by the controller.
It is easily seen that the standard test of an audio rendering system made according to this embodiment will result in a measurement of noise at a very low level. As above, the usual test is to apply a 1 kHz tone at −60 dB relative to full scale, and measure the total noise. An audio rendering system using the described embodiment will recognize a that the signal is only at a frequency of 1 kHz, and might adjust high pass filter 202 and low pass filter 204 to pass, for example, 900 Hz to 1.1 kHz.
Consequently, the noise bandwidth would only be 200 Hz rather than the standard 20 kHz; using the formula above, 5 nV times the square root of 200 Hz would equal 70.7 nV RMS. This is one-tenth of the 707 nV RMS noise calculated above for a current state of the art audio rendering system made with the same components (other than the adjustable components).
One might regard the described embodiment as enabling an advanced form of auto mute; rather than shutting off the entire channel only when silence is detected, an audio rendering system made or operating in this fashion may mask noise in those bands that are not currently carrying any signal and allow noise only in bands that are carrying signal information.
At step 1004, the digital audio signal is converted to an analog audio signal, for example, by a DAC.
Next, at step 1006, a bandwidth sufficient to contain all of the frequencies found in the digital audio signal is determined. From this, cutoff frequencies for high pass and low pass filters are determined. These determinations may again be done by the controller.
At step 1008, the values of adjustable components in the high pass and low pass filter necessary to result in the determined cutoff frequencies are determined, again, for example, by the controller.
Finally, at step 1012, the analog audio signal is passed through the high pass and low pass filters, limiting the analog audio signal to the determined bandwidth. As described above, this will limit the noise in the output analog signal. The method then returns to step 1002 to process the audio signal in the next time interval.
In this way, the described circuit and method limits an analog audio signal to a bandwidth that contains all of the frequencies present in the audio signal, without always requiring that the entire audio spectrum of 20 Hz to 20 kHz be present in the audio signal. Noise in the analog signal is thus reduced. As described herein, and further in the '617 application, certain types of adjustable circuits will result in additional advantages, in the form of reduced waste of power.
The disclosed system and method has been explained above with reference to several embodiments. Other embodiments will be apparent to those skilled in the art in light of this disclosure. Certain aspects of the described method and apparatus may readily be implemented using hardware configurations or method steps other than those described in the embodiments above, or in conjunction with elements other than or in addition to those described above.
While the embodiments described are particularly applicable to a smartphone or other mobile device, other non-mobile audio applications may also benefit from the application of the circuits and methods described herein. Thus, while the method steps described herein may, for example, be performed by a dedicated controller as described herein, one of skill in the art will appreciate that a general purpose computer or processor, whether within a mobile device or non-mobile computer, may also be used.
These and other variations upon the embodiments are intended to be covered by the present disclosure, which is limited only by the appended claims.
This application claims priority from Provisional Applications Nos. 62/249,483 and 62/249,490, both filed Nov. 2, 2015, which are incorporated by reference herein in their entirety.
| Number | Date | Country | |
|---|---|---|---|
| 62249483 | Nov 2015 | US | |
| 62249490 | Nov 2015 | US |
